LinkedIn introduces a “Report AI Slop” button. 

Well this was inevitable.

LinkedIn has introduced a “Seems like AI slop” feedback option, allowing users to flag posts and comments they consider generic, low-value or lacking in substance. It forms part of a wider effort by the platform to reduce repetitive, automated and poor-quality content appearing in feeds, alongside improved classifiers designed to identify it and changes to how widely that content is distributed.

People haven’t exactly been shy about using it either, according to LinkedIn Chief Product Officer Hari Srinivasan, more than one million people used the feedback option within its first two weeks, while LinkedIn reported that posts classified as AI slop were receiving around 40% fewer views than they had just a few weeks previously.

For anyone responsible for a recruitment company’s social media strategy, that should be worth paying attention to. This is obviously a much wider conversation than recruitment, but our industry has become a particularly good demonstration of how quickly something can go from looking new and innovative to being absolutely everywhere and, as a consequence, remarkably easy to ignore.

We’ve watched the slopification of recruitment marketing happen in real time

At Halt, we spend our working lives around recruitment brands, websites, content and social media, so we inevitably see an enormous amount of recruitment marketing every week. Over the past year or so, we’ve noticed a fairly significant change in how some businesses are approaching their content as generative AI has made it possible to produce increasingly sophisticated copy and creative in seconds.

Some recruitment businesses that previously relied on an in-house marketer, designer or specialist outsourced recruitment marketing agency like Halt have started moving towards an AI-first approach. I can understand why it is tempting, as businesses can theoretically remove the cost of somebody spending hours understanding an audience, learning the nuances of a brand, developing a tone of voice, thinking about the message, designing the creative and deciding how all of that should come together, replacing much of the process with a prompt and something that appears on screen within a few moments.

To give the technology some credit, the first impression can be pretty convincing, and I’ve developed a slightly unhealthy habit of spotting AI-generated recruitment graphics in the wild, and once you start looking for them, they’re difficult to unsee. They’re often glossy, detailed and visually dramatic, with city skylines, server racks, industrial pipework, enormous flags, glowing icons, salary figures, benefits, calls to action and motivational straplines squeezed into whichever corner of the image hasn’t already been occupied.

Somewhere along the way, the humble recruitment graphic appears to have decided that it also needs to be the job description, careers page, company brochure and employer value proposition, all in one image.

One of the examples we’ve seen even displays LinkedIn’s Content Credentials information identifying OpenAI’s Media Service API as its creator. That credential doesn’t mean LinkedIn has classified the image as AI slop, content credentials provide information about how media was created or modified, whereas the AI-slop feedback system is concerned with generic, repetitive and low-value content.

For me, the interesting part is what happens when more businesses start handing the same creative decisions to the same technology, because we’re already beginning to see the results.

Your logo isn’t your brand

Recruitment businesses have invested serious amounts in branding over the past decade, the days when changing a logo, choosing a shade of blue that looks like Facebook or the old Twitter brand to constitute a rebrand are thankfully largely behind us.

A good identity is built through hundreds of deliberate decisions around typography, photography, composition, spacing, illustration, tone of voice, messaging, hierarchy, movement and personality. When those things are applied consistently, they create recognition, and recognition is incredibly valuable when your business is competing for attention in the same LinkedIn feed as thousands of recruiters talking to the same clients and candidates about many of the same subjects.

The problem with asking a generative model to “create a professional recruitment graphic using our brand colours” is that most of those decisions have suddenly been handed to the model. Your logo might be present and the colour palette might be close, but the typography, composition, imagery, hierarchy and overall creative direction can bear very little resemblance to the identity you’ve spent years developing.

Repeat that often enough and the brand gradually becomes whatever the AI tool thinks recruitment marketing should look like, which would be less concerning if it didn’t appear to think everybody should look remarkably similar.

For businesses that have invested thousands, developing distinctive brands, websites and creative systems, slowly replacing that identity with generic AI creative seems a rather efficient way of undoing some very expensive work.

LinkedIn is becoming one enormous game of spot the difference

This is perhaps the part I find most interesting, Generative AI gives us access to an almost unlimited number of creative possibilities, yet we’re somehow using that technology to create an extraordinary amount of content that looks the same.

Scroll through enough recruitment posts and a familiar formula starts appearing with cinematic backgrounds, oversized headline, glowing accent colour, neat little row of icons, blocks of supporting copy, enormous salary figure and perhaps a handwritten motivational phrase for that final flourish of authenticity.

Put five of these posts next to each other and the fact that one is green, another purple and another blue doesn’t suddenly make them distinctive brands. They begin to look like different skins applied to the same design system.

There is also the amount of information being forced onto individual graphics. Social creative should give somebody a reason to stop scrolling and find out more, but we’re increasingly seeing images attempting to communicate every requirement, benefit, qualification, piece of software, project type, location, salary, contact detail and company proposition simultaneously. If your LinkedIn graphic needs a zoom function, we may have gone too far.

LinkedIn itself appears to recognise the wider issue. Its guidance describes AI slop as low-effort, likely AI-generated content that may look polished while lacking a clear point of view, unique perspective or substance, and the definition extends beyond writing to content that feels generic, repetitive, recycled or primarily engineered to attract attention.

This definition is worth considering carefully, whether somebody can now prove that an AI tool was involved is almost beside the point, what matters increasingly is whether the finished content looks, sounds and feels like your business, or whether it could belong to virtually anybody with access to the same software.

The brand implications are concerning, the reach implications are worse

Recruitment leaders could reasonably dismiss some of this as marketing agencies and designers getting slightly precious about creative standards, we are, as we’re the  are people who can lose half an afternoon discussing whether something needs another six pixels of spacing.

The LinkedIn changes make the conversation rather more commercial, they’ve been clear that its focus is on the value content provides, and AI-assisted content can still perform perfectly well when it contains genuine perspective, expertise or experience. The platform is investing in systems designed to identify generic and low-quality content, particularly when deciding what should be recommended beyond somebody’s immediate network.

That creates a potentially uncomfortable equation for businesses that have embraced AI primarily because it allows them to produce significantly more social content for significantly less money. If that additional output becomes increasingly indistinguishable from everything else in the feed, audiences become less inclined to engage with it and LinkedIn becomes less inclined to distribute it, the efficiency gain starts looking rather less efficient.

You can produce ten times as much content, but ten times as much of something nobody particularly wants to see isn’t necessarily a marketing strategy.

For the avoidance of doubt, I’m very positive about AI

We’ve introduced AI into several of our internal processes and it has made our talented designers, developers and marketers considerably more productive. We can accelerate research, analyse large volumes of information, interrogate datasets, structure ideas, identify patterns, automate repetitive tasks and remove some of the administrative work that nobody chose a career in marketing because they were desperate to spend Friday afternoon doing.

Used carefully and intelligently, I think it’s brilliant, and we’re probably still scratching the surface of what it will allow good marketers, designers and developers to achieve.

LinkedIn appears to take a similar position, its own guidance makes clear that AI-assisted content is welcome when it reflects genuine perspective, experience and expertise, and the platform recognises legitimate uses including refining language, proofreading and helping people communicate their ideas more effectively.

The question I think businesses need to ask is how far down the creative process they are prepared to hand over control.

Why we introduced an AI policy at Halt

We’ve thought about this enough at Halt that we introduced our own AI policy covering how we use the technology and, perhaps more importantly, how we want clients, partners and collaborators to communicate with us.

We ask people working with us to send us their actual thoughts, reactions, feedback and opinions rather than a polished AI-generated approximation of them. Marketing and branding live in the grey areas, inconsistencies and individual characteristics that generative technology has a habit of politely tidying away.

Give me the badly written email containing a genuinely interesting opinion. Give me the half-formed thought somebody hasn’t quite worked out how to articulate yet, the strange phrase they always use, the client conversation that annoyed them, the candidate story they can’t stop talking about, the opinion they suspect might divide people or the slightly chaotic three-minute voice note recorded from the car.

Our team can work with that, and our policy recognises that we use AI ourselves to support research, analysis and efficiency, while protecting the human thinking, creativity and communication that ultimately gives the work its value.

You can read the full Halt AI Policy here.

Recruitment has more to lose from blandness than most industries

Recruitment businesses have spent years telling clients and candidates that they are fundamentally a people business, built around relationships, judgement, trust, experience, networks and an ability to understand the things that aren’t always obvious from a CV or job description.

Against that backdrop, filling your social channels with automated content designed to demonstrate how human and knowledgeable your business is feels like a peculiar direction of travel.

The strongest recruitment brands I’ve worked with have personalities, opinions, quirks, expertise and recognisable people. Their content comes from conversations happening inside the business, things their consultants are seeing in the market, mistakes they’ve made, lessons they’ve learned and viewpoints developed through actually doing the job.

The opportunity AI gives recruitment marketers is enormous, and I genuinely believe the businesses that learn how to use it intelligently will have an advantage. It can make talented people faster, remove tedious work, unlock insights, accelerate research and give marketers and designers more time to spend on strategy, creativity and conversations.

What worries me is seeing businesses use that opportunity to produce more content rather than better content, particularly when the result gradually strips away the identity they spent years trying to build.

The cheapest way to fill a LinkedIn feed could eventually prove to be a surprisingly expensive way to dilute a brand, become indistinguishable from your competitors and damage the reach the content was supposed to generate in the first place.

And now, should your audience decide they’ve finally had enough, LinkedIn has thoughtfully provided them with a button with the word “slop” on it, which, from a marketing perspective, is fairly difficult feedback to misinterpret.